Data journalism, computer-assisted reporting and computational journalism: what’s the difference?

Is data journalism more networked and open than computer-assisted reporting (CAR) and computational journalism? The differences are examined in a journal article in Digital Journalism by Mark Coddington of the School of Journalism of the University of Texas at Austin. He has developed four dimensions in his typology, based on his analysis of about 90 texts (academic and professional) about these forms of ‘quantitative journalism’. The four dimensions, each of which he presents as a range between two opposing poles, are:

  1. professional expertise vs networked information — how far is it the limited domain of ‘professionals’ (linked also to the norms and practices of traditional ‘professional’ journalism) vs a more open, networked approach involving ‘non-professionals';
  2. transparency vs opacity — how far does it disclose the processes, practice and/or product;
  3. targeted sampling vs big data — does it gather and analyse a sample (probably then relying on inference or causality to draw conclusions) or a more comprehensive data set or collection (probably emphasising exploratory analysis and correlation); and
  4. seeing the public as active vs passive — the first linked to a more participative, interactive vision of the public, and the second to a more traditional, passive conception.

Mark Coddington’s diagram provides a useful summary of this, and how he situates CAR, data journalism and computational journalism along these four dimensions:

Typology of data-driven journalism

How Mark Coddington characterises data journalism, CAR and computational journalism. From his paper:

In some ways, the main dividing line is between CAR and the other two. This is perhaps not surprising, given that CAR has been around much longer and so — almost inherently — is tied more closely to ‘traditional’ ideas of journalism. Data journalism and computational journalism, on this analysis, have more in common, but perhaps differ most clearly in two ways. Data journalism is characterised as more ‘open’ (transparent) than computational journalism, and as less ‘professional’ in its orientation — ie more networked and accessible to those who are not ‘professional journalists’. (Data journalism as the new punk, anyone?)

Most data journalists (plus CA reporters and computational journalists etc) are unlikely to be bothered by how their work is classified, as Mark Coddington notes — mentioning Adrian Holvaty’s “Is data journalism? — Who cares?” post. But it does matter to researchers. Why? Because, he explains, “these definitional questions are fundamental to analyzing these practices as sites of professional and cultural meaning, without which it is difficult for a coherent body of scholarship to be built”.

He adds that this is an initial attempt at classifying CAR, data journalism and computational journalism, in what is still an emerging and developing field. Also, his study relies heavily on research in the USA and Scandinavia. While much of his typology rings true to what I know of data journalism in the UK (and CAR and computational journalism, to a lesser extent), I wonder how far it might differ here, and indeed elsewhere.

My interest (apart from running an MA programme that includes data journalism) stems partly from having written about the development of data journalism in the UK in a chapter in Data Journalism: mapping the future, That is when I came to realise how far the emergence of data journalism in the UK drew on US journalism’s experience of CAR, trainers from the States etc — helped along by the arrival here of the Freedom of Information Act and the open data movement. I’ve also touched on this topic in discussion with a US journalist who said he saw not difference between CAR and data journalism.