The Task at Hand: Causes for Concern and Reasons for Hope

With data coursework lacking in so many schools, the strongest presence of data journalism in most of academia has been the study of changing newsrooms by sociologists and communication scholars. Their work aims to document and explain data practices within ongoing scholarly conversations about media, technology, information, and society.

Elsewhere in academia, narrative uses of data and computation have emerged independently. Besides the work of quantitative social scientists, like those who inspired the work of Meyer, significant movements in the arts and humanities treat data either as a novel inroad to their traditional objectives or as a means to reinterpret those objectives. Probably the broadest of these movements falls under the heading of the “digital humanities.” One of its leading figures, Franco Moretti of Stanford University’s English department, has developed methods of “distant reading” by which one asks questions of a set of books larger than any one person could read in a lifetime. Dennis Tenen, a professor in Columbia University’s English and Comparative Literature department who has also taught at the Journalism School, identifies himself as a practitioner of computational cultural studies and argues that most disciplines have by now developed computational methods that have either complemented or supplanted their earlier practices.⁶

Several universities have founded centers and institutes devoted to work at the nexus of data, computation, and humanistic endeavors. The University of Illinois, Urbana-Champaign, for instance, hosts the Institute for Computing in Humanities, Arts, and Social Sciences, or I-CHASS, a partnership between the university and the National Center for Supercomputing Applications. The institute helps develop partnerships among social scientists and computing experts, engineers, data scientists, and computer scientists. Their collaborations have included work on large-scale video analysis, research into climate change, and even digitizing and analyzing the papers of Abraham Lincoln.

The uses of data and computation in architecture, geography, and economics also reflect the manner in which these disciplines adopted new tools and methods in recent decades. In journalism, our history is not so different. Like data journalism, computational work in the humanities and social sciences is growing, and this is reflected in the relatively healthy academic job market for digital humanists compared with the job market for traditional scholars.

Overall, we see data science and computational methods being introduced into disciplines across universities that, like journalism, have not been particularly quantitative in the past. Practices involving the use of data and computational methods may be bundled into entirely new departments, centers, research institutes, and degree programs (such as data science and computational media). It is not the purpose of a program in data journalism to compete with these other disciplines, but to develop a curriculum that is intrinsically journalistic—one that reflects a mission to find and tell stories in the public interest—as well as develop partnerships and collaborations with other disciplines.

One example of unexpected interdepartmental collaboration at Columbia has been with the Earth Institute, which has curated a massive database of climate data and offers courses in Python programming in which several Journalism students have enrolled. This course focuses on large time-series data sets, which enables data journalists to put the climate into context in their stories.

In 2013, Jean Folkerts, John Maxwell Hamilton, and Nicholas Lemann—all journalism school deans and two of the three of them longtime professional journalists—published “Educating Journalists: A New Plea for the University Tradition.” The paper focused on “universities’ role in journalism as a profession” but it also discussed how this transformation in journalism could be a boon for the schools that educate journalists. The authors wrote:

That journalism is going through profound changes does not vitiate—in fact, it enhances—the importance of journalism schools’ becoming more fully participant in the university project. Done properly, that will produce many benefits for the profession at a critical time. Journalism schools should be oriented toward the future of the profession as well as the present, and they should not be content merely to train their students in prevailing entry-level newsroom practices.⁷

Key among their recommendations was this: “We see all three of these early strains in journalism education—practice-oriented, subject matter-oriented, and research-oriented—as essential. And all of them can and should be applied, with potentially rich results, to the digital revolution. Journalism schools should embrace all three, not choose one and reject the others.”⁸

Journalism programs, with their ability to communicate to a general audience and their potential to analyze and visualize data for story, are a perfect partner for other departments. For example, at Stanford’s new Computational Journalism Lab (co-founded by one of this report’s authors), faculty are working on several projects with professors from other academic disciplines whose research mission touches on the same data. One goal is that data sets can be collected, analyzed, and used in academic research as well as for journalistic storytelling. In some instances, new methods of analysis can be developed in concert with important public accountability journalism projects.

Talk to deans of journalism schools today and you will hear the same refrain and the belief that data journalism, while not a savior, is an increasingly important component of how journalism education can evolve.

Steve Coll, the dean of the Columbia Graduate School of Journalism, describes the emergence of instruction in data-driven reporting practices as a recognition that data journalism is about more than just publishing stories through digital media, but about developing reporting methods appropriate to the complexity of the world today.

“Data journalism and tools like sensors look powerful because, in comparison to the way journalism schools have responded to previous iterations of technological change, this one runs deep, and to the heart of professional practice. It’s not about shifting distribution channels, or shifting structures of audience,” Coll said. “It was very tempting, in many ways necessary, for journalism schools to rush over to the teaching of tools, the teaching of platforms, the teaching of changing audience structure. But that transformation often had little to do with the core, enduring purpose of journalism, which is to discover, illuminate, hold power to account, explain, illustrate.”

Journalism schools, by necessity, adapted many new tools to respond to the massive and rapid shift to digital media. But delving into data journalism brings journalism back to its journalistic mission and moves it ahead in its research mission at the same time, Coll said.

“What we’re really seeing now is that this is a durable change in the structure of information, and therefore a need to durably change a journalist’s knowledge in order to carry out their core democratic function. Not to build a business model, not to reach more people, not to have more followers, but to actually discover the truth—you need to learn this.”

The rise of data analysis may also foster cross-campus collaboration. Journalism schools, as they embrace data analysis within their already powerful ability to tell stories, are uniquely suited to be robust participants and even leaders in developing means of storytelling with data.

Our research, which is focused on journalism schools, may not account for programs where data analysis is centered in another school or department that teaches this subject to students throughout the university. For undergraduates, in particular, there is little reason to offer in-house classes in subjects that students have free rein to study in another department. Yet it would require a great deal of latitude and initiative for students to construct hybrid degrees this way. Journalism students can sometimes be better served by cross-departmental initiatives that pair instructors for team teaching and connect journalism students with other disciplines that focus on data and computation. Northwestern, Stanford, Boston University, Columbia, Georgia Tech, Syracuse, and others have worked to build these interdisciplinary initiatives.

By establishing these interdepartmental bridges, schools can create pathways of collaboration between journalism, its partner disciplines of communication and media studies, and the other areas of research that share an interest in the future of technology and society.

Even as cross-departmental work increases, another challenge for journalism education will be to identify which data courses need to be framed journalistically and which others can be learned through classes framed within the methodology of other departments. In order to learn statistics, for example, students may be encouraged to register in classes offered by the math, statistics, or even political science department. The principles and objectives of these classes could apply within journalistic work, but that may not always be the case. These classes are often taught from a research or theoretical perspective. A statistics class that emphasizes survey methodology, for example, could be less useful for a journalism student.

Journalists do not often work with samples, but they do work with entire data sets. For data journalism education in particular, a more useful statistics class might be the type of instruction Meyer provided both in college courses and in IRE/NICAR boot camps, using social science to address journalistic challenges. Accommodating both techniques in a research or statistics class could foster collaboration instead of silos. In other instances, outsourcing a course may make sense. Mapping skills necessary for journalists, for example, are the same types of skills necessary for other disciplines in academia.

Yet the task of developing and adopting a data journalism curriculum comes with its own challenges. The high rate of change in digital tools, platforms, and programming languages means that there is more to teach and that classes themselves must be updated frequently. It is difficult to decipher which new techniques are just passing fads and which have the potential to remain relevant for even ten years. For this reason, it is important for classes to be designed so that they teach data and computation as fundamental styles of inquiry. Students can learn enough about the concepts behind a technique to be able to more easily learn new tools that address the technique—as opposed to focusing on the discrete tools used from time to time.

There are exceptions—the Unix command line, for example, has been as fundamental and immutable as any computing tool. This is a text-based application, still favored by developers for many tasks on Mac and Linux systems, for controlling the computer using typed commands instead of a graphical interface. And many of its core utilities remain essentially unchanged since the 1970s. Yet it is far more common to cite such examples as the ActionScript language for Adobe Flash, which was taught at several journalism schools less than a decade ago and is all but abandoned by developers today. The silver lining is that ActionScript shares many features with programming languages such as JavaScript and Python, so it may have offered a path for a student to develop other proficiencies. But it also highlights the importance of selecting techniques for journalism classes with long-term considerations in mind.

⁶. Franco Moretti, Graphs, Maps, Trees: Abstract Models for Literary History (London: Verso, 2007) and Dennis Tenen, “Blunt Instrumentalism,” in Debates in the Digital Humanities, forthcoming in 2016, University of Minnesota Press. ↩

⁷. Folkerts, Hamilton, and Lemann, “Educating Journalists,” p. 4. ↩

⁸. Ibid., p. 12. ↩