2020
RMIT University Data Journalism Pilot
Category: Innovation (small and large newsrooms)
Country/area: Australia
Organisation: News Corp Australia
Organisation size: Big
Publication date: 9 Aug 2019

Credit: Hamish Lindsay, Aaron Shankar, Peter Judd, Sonja Heydeman
Project description:
In 2019, News Corp Australia and RMIT University trialled a unique and innovative data journalism internship program. Under the supervision of senior News Corp journalists, journalism and data science students used public data sets to find significant stories that were then published using News Corp’s one-to-many model direct to their websites and newspapers.
The collaboration produced a number of articles, such as the featured story ‘Dosed Up and Dying’, published last year in newspapers and online.
The published work and positive feedback from students and industry are evidence of the pilot program’s success and it will be extended this year.
Impact reached:
This initiative generated articles, including ‘Dosed up and Dying’, which raised public awareness of prescription opioid use in Australia and provided, for the first time, suburb-level profiles of prescription rates, correlated with many other datasets.
It was significant for industry, allowing News Corp Australia to help develop crucial and necessary skills in journalism students and their own staff. The initiative brought new talent into the newsroom and helped introduce data science students to a career path they may not have previously seriously considered. This in turn potentially enriches the industry further.
For RMIT, journalism program educators are on the cutting edge of industry requirements and use this information to better equip the next generation of journalists. It provides students with experience in cross-disciplinary collaboration and the internships give students an invaluable insight into and experience in using these highly sought-after skills while working with senior journalists.
RMIT’s Associate Dean of Computer Science and IT Professor Lawrence Cavedon, welcomed the opportunity for students to intern with such a large organisation and work with high-quality journalists, and students from another discipline. “These internships provided incredibly valuable experience, and one different and arguably richer than many other internship opportunities.”
‘Dosed up and Dying’ was the first step in an initiative that will have far-reaching implications for the development of both journalism and data science students at RMIT.
Techniques/technologies used:
The team, Hamish Lindsay (journalist) and Aaron Shankar (data scientist), downloaded readily available opioid prescription data from the Australian Institute of Health and Welfare and in subsequent weeks discovered systemic higher prescription rates in regional communities.
The suburb level prescription data used conformed to the Australian Bureau of Statistics SA3 standard, which enabled correlation of many other external datasets.
The team first reclassified every suburb whether it was located in a State capital or a regional population centre.
Shankar said the RMIT pilot found correlations “between prescription opioid rates in an area and the cancer incidence rates, socioeconomic status, arthritis and osteoarthritis rates and the median age of the population”.
“Regression and dominance analysis was conducted in order to gauge the relative amounts that these variables contribute to prescription opioid rates. This resulted in a number of potentially problematic areas to be highlighted including overprescribed areas and areas with a potential lack of access to adequate healthcare.
“The Python programming language was predominately used for data pre-processing, analysis, and visualisations tasks, while R Studio was utilised for a regression analysis. Simple spreadsheet tasks were completed in Microsoft Excel.”
Shankar and Lindsay produced dozens of visualisations to explore the data using the Seaborn library, swarm plots most effectively capturing the underlying story of difference.
Shankr also subsequently produced a research report, “Data Journalism Pilot – An Exploration Into Data Science in Journalism”, in which he recommended Australian media organisations do more to incorporate data journalism into newsroom workflows.
The pilot produced granular insights for every town and suburb in Australia, including relationships between:
Opioid prescription rates, net prescriptions, scripts per minute;
Median income;
Unemployment rates;
Remoteness;
Socio-economic quintiles;
Cancer rates and incidence;
Mortality rates;
Age distribution;
Musculoskeletal Diseases;
Arthritis numbers and ratios; and
Aged care beds, ratios per 100,000.
What was the hardest part of this project?
News Corp Australia Newsroom Operations Manager Peter Judd said the biggest challenge was “selling” in the concept of student hybrid data journalism teams to traditional newsrooms – and getting them to trust the results.
The RMIT pilot is effectively a disrupter, a cross-disciplinary approach where the schools of data science and journalism collaborate to tell stories embedded in data, then work with a network of senior newsroom journalists to make the story public.
The reality is that these skills do not exist in many newsrooms, so editors are cautious about “going out on a limb” in a leap of faith on something they struggle to validate.
This is part of a bigger problem where journalists, who pride themselves on speaking truth to power and holding institutions to account, are powerless to do so if they can’t interrogate the information themselves, but rely on the analysis of others.
The reality is that journalists will not become data scientists and data scientists will not become storytellers by confirming their narrative and talking to people.
In our pilot workflow, the data scientist finds the story lead, the outlier, the anomaly in the data, and the journalist then prosecutes the investigation, armed with unprecedented insights that often surpass the knowledge of experts in the community.
The issue of trust was also compounded by traditional community wariness of student journalism, making it much harder to secure the interviews and the time necessary to check the story leads on the gorund.
Overall, the challenge rested with merging the needs of the country’s biggest media organisation and the senior journalists with the academic requirements for students. It also required a complete rethink of the university’s more usual Work Integrated Learning process and a commitment to genuine collaboration.
What can others learn from this project?
This project demonstrates the mutual benefits of strategic industry and interdisciplinary partnerships.
While journalism schools use many techniques to teach data journalism, research continues to call on educators to look at more inventive ways to cultivate skills.
The current literature shows there is much work being done by universities to bridge the gulf between journalism schools and computer science departments through the creation of research centres, developing joint degrees and co-teaching.
Some have also successfully used project-based assessments and industry collaboration to develop data journalism skills.
But the literature, including Weiss & Retis-Rivas, also calls for more research into innovative pedagogical practice.
Given universities are facing challenges producing graduates with data journalism skills, this conceptually innovative project strategically uses Work Integrated Learning as a vehicle to address current and future industry needs.
Journalism and data science academics consistently collaborated, working closely to nominate and select candidates – monitoring the process through feedback and assessments.
This has been combined with the extraordinary commitment of News Corp Australia to work with students to develop important stories along with the skillsets of participants with demonstrable results.
News Corp Australia Newsroom Operations Manager, Peter Judd has worked on-site with student teams based at RMIT.
In 2020, apart from increased student numbers, additional opportunities potentially include on the ground research and reporting.
This pilot has great significance broadly as a potentially useful tool that could be adopted by other universities who team with industry partners.
RMIT and News Corp Australia would be delighted to share details of this pilot with the international data journalism community.
Project links:
www.dropbox.com/sh/qiun0c3r6uacrhl/AADyY4bEP9DWROpo2gYeuJhCa?dl=0