One group visualizes when it is best to convey data findings over radio instead of charts depending on the audience’s data literacy and reliance on traditional media for information.

Data journalism has tremendous potential to drive transparency and reveal corruption in developing countries and many donors are funding data journalism as a means to good governance and transparency. The Defining and Designing Successful Data Journalism Initiatives in Developing Countries session at the 2014 Open Knowledge Festival focused on what it takes to grow a sustainable data community. In groups composed of journalists, developers, CSO representatives and other open advocates, we evaluate the most common strategies such as conferences, boot camps, fellowships, hackathons and reporting grants and discussed openly whether they had produced concrete data journalism that has had a social and policy impact.

To kick things off, I shared my own perspective on successes and failures by the media development community. The fastest and cheapest way to try to introduce data journalism is through Data Journalism Boot Camps. The model is based on the belief that a week-long (or sometimes three-day) training in the fundamentals of data journalism would give journalists just the push they needed to start fighting corruption through awesome visuals and news apps. But, just as a one-week conventional boot camp can hardly be expected to produce a special ops unit, the workshops do not produce a cadre of global journo-coders and what’s worse, they often gloss over topics like privacy, statistical errors and ethical reporting. In fact, many boot camps never result in a single data-driven story, despite a flurry of Twitter traffic that suggested that places like Nigeria, Nepal, and Bolivia are the next rising stars in data journalism.

My contrasting success story explored the four-month fellowship model where a group of talented media professionals (journalists, a graphic designer and developer) in Kenya dedicated themselves to learning new skills and producing experimental content and storytelling for their home media outlets but working from the Internews in Kenya data desk. For us, the fellowship served several purposes. It immersed the fellows in months of intensive training as each module built up their cumulative skills; it introduced them to data sources, including think tanks and government, that the journalists could use for stories when they returned to their media outlets; and it encouraged them to work as a team to complete two major investigations each, which they published before the end of the fellowship and earned them a reputation as data journalists.

Next, the 50+ participants worked in groups to identify their most illustrative examples of success and failures to help others design smarter activities. Examples included great budget visualizations that engaged the public through the media, but only one person on the team actually knew how to use the software to produce the visualization. On the flip side, another group had created a website to visualize aid data but nobody ever accessed the site and those who did found it confusing. Both of these experiences illustrated the need to establish who needs to be trained to ensure a viable product that can be maintained after the initial launch and how the target audience consumes news, which, in the case of many of our participants, is offline.

A common theme addressed difficulties in bringing together developers and journalists on projects. One group sited a successful example of a mapping platform in Latin America that superimposes environmental journalism stories on maps of environmental data such as protected zones and mining areas. This way, both data and traditional storytelling are present on the same platform. Another group paired an NGO that designed and implemented a public opinion poll with journalists who published the results. In both cases, the developers and statisticians generating data stuck with the skills they were good at while the journalists became more data literate but were not expected to become journo-coders in order to report on data.

Another participant from South Africa highlighted the challenges of embedding coders in newsrooms, who generally end up either isolated or overwhelmed by newsrooms either indifferent to data or too demanding for digital products. This experience echoes embedded coder challenges faced by a similar program in Kenya. Overall, participants shared honestly and openly about successes and failures and advocated for more “Fail Faire” type events where practitioners share knowledge and experiences not only with each other but also with donors.

These were the overall conclusions:

1. Sustainable data journalism activities require the buy-in of journalists, developers, editors and publishers
2. Finding good matches between media outlets, CSOs and developers all committed to data are key to productive collaborations
3. How people consume information should dictate narrative or visual form of data products
4. Data journalism requires teamwork, whether inside or outside of the newsroom
5. Mentoring and consulting data experts can help avert mistakes in data analysis and interpretation
6. Storytelling to convey data helps people understand and connect with the issue
7. Not a lot of resources are available for data journalism tailored to developing country contexts
8. Data integrity is an emerging issue of concern as data journalism increases in popularity
9. Topics, projects or specific production goals can help make data journalism activities more realistic and achievable
10. Sharing lessons learned is essential to designing more effective data journalism activities

To learn more about how to design data journalism initiatives in developing countries, check out the session Etherpad and my blog post for Knight Mozilla Open News Source on Developing Data Journalists in the Developing World.

Thirteen Afghan journalists participated in the first-ever data journalism workshop held in Afghanistan. (credit: Internews)

During a training workshop held by Internews, Afghan journalists immediately saw the potential of data journalism to combat corruption, and wanted to jump right into ambitious investigative topics such as tracking foreign aid budgets and political and business interests in the mining sector.

The five-day, hands-on workshop, held in Kabul and supported by the USAID-funded Afghan Civic Engagement Project (ACEP), was Afghanistan’s first ever data journalism training.

Internews trainer Eva Constantaras introduced 13 senior journalists to skills such as finding public-interest angles in government and NGO data, analyzing election results using Excel, and visualizing data with Google Charts. The training marks a first step towards a more advanced, data-driven media sector in Afghanistan that facilitates greater governmental transparency and accountability. While the training focused specifically on election data, the skills are widely applicable to coverage of a range of pressing development issues in Afghanistan. Over the course of the week in March 2014, trainees explored health, economic, education, and gender data through analysis and visualizations.

“Data journalism can be quite daunting for journalists because it requires such a wide range of skills, a lot of practice and teamwork,” Constantaras explained. “These trainees from day one developed a list of investigative stories they wanted to pursue through data, which provided motivation as they struggled through a lot of the more technical skills such as writing formulas in Excel and choosing appropriate graphical forms for different data types.”

The workshop inspired one journalist in particular, Rohullah Armaan Darwesh, from the Payk Investigative Journalism Centre. Using an ACEP media civic engagement grant, Payk delivered a series of investigative reports on elections-related issues. As part of an article in this series on the rising cost of weapons in the build up to the April elections, Darwesh utilized his new skills to visualize the 30% rise in gun prices and the importance of security as an election issue for Afghans.

“Data journalism helps journalists show and explain their research and reports in a better way in less time,” said Darwesh. “I learned that a reporter can [use comparative charts] without the support of a graphic designer… My work used skills I learned from the data journalism training workshop.”

While available data is currently limited in Afghanistan, it is hoped that data journalism and the public’s demand for investigative reporting will accelerate the open data movement there.

The data journalism fellows

This guest post was written by Eva Constantaras and Jimmy Makhulo, and it was originally published on Internews in Kenya.

Five Kenyan media professionals, including two print journalists, a TV journalist, a developer, and a graphic designer, graduated as the first class of Internews in Kenya Data Journalism.

On February 21, 2014, the fellows celebrated after completing the sixteen-week data journalism training and production that raised awareness about the misspending, corruption, and inequality that plague Kenya’s public healthcare system. During the fellowship, the participants learned how to access, scrape, analyze, and visualize data using digital tools. They also gained an appreciation of interconnections in data — with the ultimate aim of unearthing stories buried in data through investigative journalism.

“The more we worked in data journalism, the more we realized that data journalism is teamwork,” explained Dorothy Otieno, the Internews in Kenya Data Journalism Trainer who designed the fellowship to build a sense of community.

The fellows’ stories included reporting on issues such as malaria, family planning, maternal and child health, and HIV.

Health financing in Kenya

Driven by too many questions on health financing in a devolved system of government, fellow Paul Wafula, a journalist from The Standard Investigative, used his data journalism training to get answers. His story Health for Change was published as a five-part series in The Standard.

“I have come to appreciate what we have been missing in our reporting because the story I have done, if I had done it without data, my email would be full of questions,” says Wafula, who examined whether the county budgets reflected local health crises.

“We expected that under devolution, the government would respond to the basic needs of the people, and in West Pokot, they would now have the resources to deal with eye issues, or if you are in Kilifi, you could address elephantitis because that is your most pressing health issue that had been neglected by the national government. But I discovered that the county governments were falling back on the national government’s way of doing things.”

Watch this video to find out how he did the cross-platform multimedia story:

Last month, the fellows saw some of the first concrete policy results of their investigations as they shifted from data as evidence to data that reveals the root of the problem of poor healthcare in Kenya.

Holes in the national safety net program

Wafula’s second investigation focused on an emergency cash transfer safety net program for the poorest Kenyans, including orphans and vulnerable children, and revealed that legislators had changed the distribution formula for allocating Sh12.3 billion ($142 million) from an earlier model in which government poverty indicators dictated distribution to an equal division among counties in order to win the favor of wealthy constituencies.

“Being a poverty program, I knew that the poorer the place, the larger the allocation,” explained Wafula. “Here you found that a constituency in which 30 per cent of the constituency was poor was getting the same amount as a county where 80 per cent of the population was poor. I realized that if you were poor in Lamu, you were eight times more likely to get money than a poor person in Turkana.”

After the story ran, Labor cabinet secretary Kazungu Kambi revealed to Wafula that Sh600 million ($7 million) were handed out to ghost recipients. “I knew money was disappearing and was given to people who were not deserving, but I didn’t know how much or where it was going,” explained Wafula, who doesn’t believe that the government would have admitted to the corruption if it were not for his initial investigation of the distribution failure. Kambi has ordered an audit to identify and remove ghost recipients and other undeserving cases, developed new vetting committees that include community leaders to identify recipients, and enlisted a private telecommunications company, Safaricom, to distribute the funding.

Several agencies funding the project have raised questions about why the government allowed politicians to change the original distribution plan that had been approved by donors. Various stakeholders contacted the journalist and requested his raw data for further analysis.

Linking climate change and malnutrition

When the sun sets in Turkana: Hunger stakes and stripes in the North by fellow Mercy Juma, a broadcast journalist at NTV, ran as the lead news story on January 21. The twelve-minute data-driven story reveals that malnutrition in children is a growing problem in Kenya as famines become more intense and frequent.

The station phone was ringing off the hook before the story finished airing. Due to the massive reaction to the story from individuals and organizations, within hours the station established a relief fund for Turkana County, as explained in the follow-up story Famine strikes again, which brought in more donations.

Juma followed up the article with a TV story, Hunger keeps children away from school in Turkana. Since then, there have been prominent stories on the desperate famine situation in Northern Kenya daily in the Kenyan media, which has historically shown a lack of interest in the plight of the isolated and impoverished regions of northern Kenya.

Even more important for the fellows who worked on the story, the Drought Monitoring Committee asked Juma to share data from her story because they claimed they were not aware that the situation had become so desperate, though the same department had tried to charge her for access to the data when she began her investigation.

Based on Juma’s water shortage data, the Ministry of Water plans to travel to Turkana to dig more boreholes. The government, through the Ministry of Planning and Devolution, released Sh2.3 billion ($27 million) to go towards relief distribution in Turkana County, a development that Juma is following closely.

An in-depth look at contraceptives

Fellow Mercy Juma also produced a cross-platform multimedia story that links the high number of unsafe abortions in Kenya to low contraceptive use. Her ten-minute TV story Grave Choices was aired on NTV. The print version was published as a cover story in The Daily Nation. In this video, Juma explains how she looked deep into data to get insights on the issue of unsafe abortion.

A parallel story by fellow Samuel Otieno illustrated the financial drain on the country caused by fixing botched abortions rather than investing in contraceptive access.

Data journalism beyond the fellowship

Since data journalism is still in its infancy in Kenya, the fellows now have established beats that give them more freedom to pursue data-driven stories about health spending, drought, poverty, and other key governance issues when they return to their media outlets.

They presented their stories at the inaugural Online News Association Nairobi event, discussing their process with other digital journalists in Kenya and growing the data journalism community. “It was an interesting journey that we were required to develop a culture of teamwork right from the story conception phase,” said Michael Mosota, the graphic design fellow from The Nation. “I have picked up that culture of teamwork and now all the stories I work on I participate from the conception stage and I train other designers on print and online visualization tools.”

Dorothy Otieno explained the attitude shift at the media management level at The Nation. “It wasn’t me calling the editors to tell them about data journalism,” she said. “It was an editor calling me and me telling him that he could read the story in his paper on Tuesday because another editor had agreed to run it.” The audience had seen many of the stories on television and in the newspaper and expressed surprise at the policy impact that the young journalists had made in such a short time. They expressed that Kenyans want to see money allocated in the right place and health issues addressed through policies that are grounded in data.

The fellows are trailblazers in the developing world, which has so far escaped the media industry crisis that has forced many Western media outlets to explore data journalism as a strategy to become solvent in the age of free online content. “You should be commended for transforming yourselves into data journalists not to keep your jobs but rather because you are passionate about journalism and convinced that your data-driven stories can help Kenyans become more engaged citizens and influence the direction of this country,” Data Journalism Advisor Eva Constantaras told the fellows at the conclusion of the ceremony.

For more information, see the stories produced by the Data Journalism Fellows on Data Dredger.

This is a cross post from the Internews Kenya Data Journalism Fellowship Blog.

Internews in Kenya’s third week-long data journalism workshop, which wrapped up on August 24, confirmed our growing suspicion that in a week, aspiring data journalists can master the fundamentals of data journalism, but it isn’t easy. From finding and scraping data to cleaning, analyzing and visualizing data, it’s a crash course in both the theory and practice of applied data science. A carefully planned and balanced training program is designed to whet journalists’ appetite for more and open up access to the data journalism community.

###1. Find good storytellers
Data journalism transforms journalists’ role from information transmission to information creation. Selecting journalists who have already demonstrated critical thinking and analytical skills through investigative or feature reporting is essential. No amount of data or software can replace effective narrative that explains the human implications of the data. Choosing participants with a proven nose for news saves from having to explain why data is important.

###2. Develop a cohesive training program
Data journalism requires a diverse toolkit even at the basic level, and honing in on one topic and using the same datasets for all the exercise helps trainees understand how each step and each tool serve to dig deeper into that topic. So if the topic is education, the exercises for advanced site searches, scraping PDFs or html and analyzing data through Excel should all focus on publically available education data. The number of tools and skills can seem overwhelming and a consistent topic will help trainees undertake a data-driven investigation from beginning to end.

###3. Don’t leave anyone behind
Data journalism isn’t for everyone, but it should be for everyone in the workshop. Having enough trainees on hand for individual attention helps ensure that all trainees are following each step of the activities and that the class doesn’t move on until everyone has got it. Trainees will be even more committed if they understand that if they don’t come out of the data cleaning session with a clean dataset, they won’t have material to analyze in the next session or to visualize in the session after that.

###4. Show and tell
People have different learning styles and while providing background materials and resources can help, presenting the information in different ways during the sessions will smooth the path for many. Providing the full presentation with easy-to-follow, step-by-step instructions and screen shots to teach new software will compliment oral presentations. If a trainee misses a step, he or she will have access to the complete presentation and can check previous slides to catch up. Providing only oral instructions also makes it likely that while the trainees all come out with the desired results during the training, they don’t fully understand how they got there or how to use the software themselves. By simply following instructions as they hear them, they may not realize how they got there or they may miss the underlying concepts.

###5. Lead them down the wrong path
It is tempting to lead trainees to fantastic data sets that reveal amazing new insights, but in the real world, many data sets don’t lead anywhere. An exercise that involves calculating the change in per capita crime rate over time may show an insignificant change. Participants may jump on ANY change as significant and exaggerate the results in their story because as journalists, the instinct is to write a splashy headline. Aspiring data journalists need to be able to differentiate between a story and noise when it comes to analyzing data and how to suppress the instinct to exaggerate findings.

###6. Simplify statistical concepts
Journalists need to know how to use data responsibly but they don’t have time for a university-level statistics class. So it is important to touch on the fundamentals of statistics pragmatically with examples of erroneous reporting that confuse gross and per capita, infer causation when there is only correlation and report dramatic changes in poll results that in reality fall within the margin of error. Through these examples the session can teach journalists how to evaluate data sources, understand the basics of statistics and the price of getting it wrong and encourage them to ask experts before jumping to conclusions and reporting data wrong.

###7. Insist on teamwork
Collaboration can be the most challenging aspects of data journalism for journalists who are used to working solo. Activities that encourage trainees to work together to brainstorm hypotheses, re-design an ineffective visualization or analyze international examples of excellent long-form data journalism serve several purposes. Team members begin to think not only in terms of playing to each members’ strengths and dividing tasks accordingly but also helps them recognize the advantage of working together to solve complex data problems.

###8. Emphasize multimedia
In newsrooms, it is often a stretch for journalists to consider putting the time aside to practice data journalism skills and telling that story through multimedia can seem even less tenable. But many of the most high impact data journalism stories are a careful balance between analysis and telling a human story that puts a face to the data through video, audio and other interactive elements. Exposure to such stories and encouragement to think of data stories in terms of a multimedia package can help journalists get out of the mindset of the daily news cycle and ready for a data journalism team.

###9. Put skills to the test in a Data Expedition
The School of Data Expedition model is an excellent wrap-up activity for participants to work as a team to employ all the skills and tools they have learned over the week. Even more importantly, it reinforces the need for collaboration to complete the expedition within a few hours and encourages teams to persevere despite the hurdles the come along with a data investigation including scarcity of data, difficulty in establishing relationships among datasets and producing digestible stories for a general audience.

###10. Create a community
Veteran data journalists rely on internal data journalism teams or external resources such as Hacks/Hackers , School of Data and Data Driven Journalism for support, inspiration and resources. If trainees engage with these global communities, they are much more likely to become local leaders in data journalism.