Review of Exploring Data Visualisations: An Analytical Framework Based on Dimensional Components of Data Artefacts in Journalism
By Cindy Royal, Professor
September 7, 2021
Recently, I reviewed the article from Journalism Practice on Preserving Data Journalism: A Systematic Literature Review (Heravi, et al., 2021) written by a group of scholars in Ireland, which discussed the challenge of archiving data journalism projects. That is why when a new article on data visualization published online on Digital Journalism also included Bahareh Heravi as an author, it caught my attention.
Exploring Data Visualisations: An Analytical Framework Based on Dimensional Components of Data Artefacts in Journalism (Stalph & Heravi, 2021) is a provocative piece proposing an improved framework for analyzing data visualizations. The authors, Florian Stalph, of the Department of Communication Studies and Media Research at Ludwig Maximilian University in Munich, Germany, and the aforementioned Heravi of the School of Information and Communication Studies, University College Dublin in Ireland, analyzed 183 visualizations, highlighting the prevalence of components and identified characteristics.
Pointing out that many research studies have focused on the appearance of visualization types, interactive features and data sources, the authors feel that these approaches address them more as editorial rather than visual products. “As important as these efforts are to further our understanding of data journalism products and what they look like, counting units and classifying data journalism appeared to hypostatize data artifacts as editorial techniques rather than diagrammatic elements” (Stalph & Heravi, 2021).
To improve this, the authors analyzed the methodological framework of previous content analyses of visualizations and then produced a synthesized framework that they applied to their dataset. The dataset included a corpus of visualizations in 78 award-winning data journalism projects via the Data Journalism Awards, spanning the years 2013-2017. The Data Journalism Awards were a project of the now defunct Global Editors Network, but have now been replaced with the Sigma Awards.
The framework consists of five components, each containing a range of dimensions: visualization type (i.e. categorical charts, hierarchical charts and news app or customized visualization, etc.), level of interactivity (six levels of increasing interactivity), data provider (government, NGO/NPO, etc.), method of access (FOI, open data, etc.) and purpose (including to inform, persuade, entertain, etc.). See the article for the full framework of dimensions.
The authors also provide a methodological framework for completing coding of visualizations that includes sample selection, recommendations for training and number of coders and analysis.
Lead author Florian Stalph, in an email interview, indicated that the strength of this framework is in how it builds on previous research on data visualization and in its reliance on theory. “In this paper, we propose a unified framework for the analysis of visualizations that draws on theorizations of data journalism practice, as well as findings from previous content analyses,” Stalph said.
Components were coded and operationalized within the dataset, producing some interesting results.
- Line charts appeared as the most commonly used visualization type accounting for 20.5% of the sample, followed by bar charts (11.9%), choropleth maps (7.6%) and flow maps (6.5%).
- Almost two thirds of the sample featured interactive functions to varying degrees. After combining levels, roughly one quarter of all visualizations were highly interactive, while 38.4% were static and 38.9% featured some interactive functions, such as the option to choose from predefined data sets via filtering or displaying overlays and annotations. Those categorized as spatial charts, like maps, had higher levels of interaction than other chart categories.
- Governmental sources represent the largest data sources (51.9%).The majority of the data used was either open access or made publicly available by data providers (66.5%).
- In terms of purpose, the most prevalent was coded as to inform (49%) with the next most prevalent being to entertain (28%).
Regarding interactivity, the authors note that this feature requires journalists to relinquish some of their “narrative control.” However, they recognized in the article that the practice offers much opportunity for audience engagement. “Interactive visualizations in particular appear to play an important role in the participatory generation of knowledge.”
Stalph identified the relevance of this research to professionals in uncovering the process behind the choices within the framework. “The components of visualizations, such as the chart type, interactive features or indicated data sources do not only impact the overall appearance,” Stalph said. “They also influence the potential of communicating journalists’ knowledge claims as they externalize the various steps of choosing a source and prescribing how data is to be read by visualizing that data in a certain way.”
This analysis provides foundational knowledge toward future research on data visualizations that includes understanding why certain visualization types are most prevalent, attention to journalistic control and the choice of data sources. The authors also recommend studying users’ perceptions of visualizations to ascertain the value they perceive from elements of the framework.
The authors broach the lack of training in statistics and data analysis and the need for more research on how journalists translate and parse data into visualizations, referencing previous research that indicated that half of journalists “had little to no formal training in analysis, statistics, coding or data science.”
However, Stalph indicated that current tools reduce barriers that previously prevented journalists from employing data in their stories.
“As tools for data visualization get more accessible, and software is being developed specifically with journalists in mind, it has probably never been easier to get into data visualization,” Stalph said. “Journalists can focus on visualizing information and zero in on diagrammatic and computational thinking.”
Stalph notes, however, that collaborations may be necessary for journalists seeking to employ more complex data sources or analyses.
“The more complex the underlying data gets and if data modeling or advanced statistical analysis is involved, collaborating with data scientists or peers trained in social science research and statistics can help to prevent misleading or inaccurate visualizations,” Stalph added.
In regard to visualization’s role in media education, Stalph thinks that all journalism students should be exposed to data concepts.
“Even those students with no intention of working as data journalists should understand the potential and limitations of working with data,” Stalph said. “At some point they might have to collaborate with a data journalist, so understanding the basics of generating visualizations out of data is essential.”
He notes, however, that there is an opportunity to improve the presence of data topics in academic programs.
“At the same time, many communication and journalism studies curricula are overlooking how strongly social science research methods and data journalism are interwoven,” Stalph added.
Personally, I think more attention should be paid to the skills within media organizations for developing data visualizations, the location within the organization where those skills exist and the power that those individuals are able to exercise. In my earliest piece on data journalism, I studied the New York Times Interactive News Technology team (Royal, 2012) and more recently have looked into the emerging role of product management (Royal, 2017; Royal 2020). As visualizations become more prevalent and increasingly more complex, demonstrating more interactive features, possibly in the form of news applications like those developed by Propublica and Texas Tribune, they will need to be managed and maintained as products. I feel this area presents some of the biggest challenge to media, as well as media education, as the mission of journalism shifts from a culture of writing and reporting to one of managing a range of digital products, visual content and interactive presentations.
In my opinion, the value of this framework for assessing data visualizations is three-fold. It provides a foundation for scholars to systematically research data journalism projects. It may give pause to professionals, educators and students to consider the choices they make when creating data visualizations. And, it is a tool that can grow as the role of data visualization evolves within media organizations. I recommend taking a look at the article to comprehend the full framework, literature review and results.
Heravi, B., Cassidy, K., Davis, E., & Harrower, N. (2021). Preserving Data Journalism: A Systematic Literature Review. Journalism Practice, 1-23.
Royal, C. (Spring, 2012) The Journalist as Programmer: A Case Study of The New York Times Interactive News Technology. https://isoj.org/wp-content/uploads/2016/10/ISOJ_Journal_V2_N1_2012_Spring.pdf.
Royal, C. (2017). Managing digital products in a newsroom context. ISOJ Journal (Vol. 7, pp. 45-66). https://isoj.org/research/managing-digital-products-in-a-newsroom-context/.
Royal,, C. (2020). The State of the News Product Community 2020. MILab Journal. https://www.masscomm.txstate.edu/media-innovation/milabjournal/newsproduct2020.html.
Stalph, F., & Heravi, B. (2021). Exploring Data Visualisations: An Analytical Framework Based on Dimensional Components of Data Artefacts in Journalism. Digital Journalism, 1-23.