The digital revolution has led to significant changes in news reporting. One is the use of social media as a tool for gathering and disseminating information, and another is the mining of data for stories.
Data journalism is hardly new. As early as the mid-1800s, Florence Nightingale used statistics to explain conditions in the British Army. But in the past decade, the rise of personal computers and an increase in open data policies - particularly by governments - has made it possible to access and manipulate huge datasets in the hunt for discrepancies and information of public interest.
Ask any serious news editor what skills they would like to see in a new recruit and you are likely to be told that, along with the fundamentals of how to spot a story, how to verify it and how to communicate it effectively and legally, a knowledge of data journalism skills is a definite advantage. And if those skills are well developed, eager young journalists may find their way straight into the investigations team.
Why is it, then, that most universities have been slow to teach these skills in sufficient depth to their journalism undergraduates? In the past year, the Centre for Investigative Journalism, a non-profit organisation with a strong reputation in data journalism training, sent out an email to nearly 20 university journalism departments in the UK offering workshops in data journalism training and got not a single reply. This would seem to reflect an ambivalence towards teaching these skills - one that isn't always shared by students, some of whom pay their own way to attend the centre's summer school and training workshops.
While some universities do teach the basic concepts and tools, it seems clear that adopting a more robust teaching of data journalism, or even basic statistics, would not only better meet industry needs in the new digital era, but boost the depth and status of university journalism courses. Without such skills, the MPs' expenses scandal would not have been unearthed, and the journalism that resulted from the WikiLeaks releases on Afghanistan and Iraq would not have been possible.
Julian Champkin, editor of Significance, a magazine that aims to make topics of statistical interest more accessible, ascribes the lack of data analysis training to a classic divide between arts and sciences in universities. Journalism is about communicating, and both journalism lecturers and students tend to be humanities-orientated. Many are not highly literate in maths or basic statistics. Champkin argues that university journalism courses could start by teaching an understanding of terms such as mean, median and standard deviation.
But as Paul Bradshaw, leader of the MA in online journalism at Birmingham City University and one of the sector's most data-literate journalists and journalism lecturers, points out that, even if journalism lecturers are literate in basic statistics, they must also be able to teach students how to manage a spreadsheet and other data software programs. Most journalism lecturers are teaching after a career in the industry that ended before the latest developments in digital data analysis and data visualisation.
There is greater scope for teaching data analysis in the small but growing number of joint honours degrees incorporating a science subject with journalism, and some MAs also now offer journalism with science or financial affairs, but otherwise journalism degrees tend to be arts-only affairs.
Some lecturers point out that data journalism is not for every student, and argue that it is better to teach just the basics. There is also evidence that students are often more concerned about learning the softer skills: managing projects, dealing with people and how to ask questions.
But there is no doubt that data analysis training is essential for those who want to contribute to the growing number of daily news stories that are the result of "looking at the data".
David Donald, data editor at the Center for Public Integrity in Washington DC, says that journalists covering the finances of a city council who cannot interpret the data are handing over control of the story - and in his view, that almost amounts to malpractice.